以 x 轴标签为日期的预测 plot

Question

I have a dataset like revenue and date.

I used arima to plot the data.

ts_data = ts(dataset$Revenue,frequency = 7)
arima.ts = auto.arima(ts_data)
pred = forecast(arima.ts,h=30)
plot(pred,xaxt="n")

When I plot the data, it produces plot like below.

My expectations are below,

I need to display values in Million for predicted values like 13.1M. I need to show x-axis as date instead of data points numbers.

I tried several links but couldn't crack it. Below are the experiments I made,

Tried with start date and end date in ts_data that also doesnt work.My start date is "2019-09-27" and end date is "2020-07-02"

tried wit axis_date in plot function that also doesnt work.

Please help me to crack the same.

Thanks a lot.

Answer 1

You can specify axis tick values and labels with axis()

plot(pred,xaxt="n", yaxt="n")  # deactivate x and y axis
yl<-seq(-5000000,15000000,by=5000000)   # position of y-axis ticks
axis(2, at=yl, label=paste(yl/1000000, "M"))  # 2 = left axis

You can specify the desired position of y axis ticks with at and the labels to be associated with label . In order to obtain the values like 10 MI have used the function paste to join the numbers with the letter M.

You could use the same method also for x-axis, even tough more efficient methods probably exist. Not having the specific data available I have used a generic spacing and labels only to give you the idea. Once you have set the desired position you can generate the sequence of dates associated with it (to generate a sequence of dates see https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/seq.Date )

axis(1, at=seq(1,40,by=1), label=seq(as.Date("2019-09-27"),as.Date("2020-07-02"),by="week")) # 1 = below axis

You can also change the format of the dates displayed with format() for example label=format(vector_of_date, "%Y-%b-%d") to have year-month(in letter)-day.